What’s Inside?
The authors test whether the income generated by security lending outweighs the potential improvement in a fund’s performance if the fund manager sells stock in response to a short-selling demand signal.
They undertake this analysis to understand why mutual funds would lend rather than sell a stock when the funds receive a short-selling demand signal from the market for that stock.
The authors find that actively managed equity funds that lend securities underperform otherwise similar funds that do not lend by a statistically and economically significant difference of around 0.5%–0.7% in risk-adjusted net returns.
How Is This Research Useful to Practitioners?
In assessing the impact of security lending on fund returns, the authors identify differences between passively and actively managed funds. Passive funds that lend securities achieve a higher return than do otherwise similar funds tracking the same index. Security lending does not affect the index’s risk-adjusted performance. Fees earned from security lending augment the returns from the index. Actively managed funds that lend securities underperform otherwise similar active funds.
Drawing on empirical research and studies, including Kaplan, Moskowitz, and Sensoy (Journal of Finance 2013), the authors argue that short sellers are better informed than fund managers. They examine why fund managers who lend securities fail to act on the demand signal they receive from short sellers by not selling the stock in question and find that the underperformance of funds that lend stocks is concentrated among funds with greater investment restrictions.
The authors also find that the restrictions are relevant only if managers are faced with short-selling demands that are systemic to their particular style. Otherwise, if a short-selling demand is stock specific, the manager can sell the highly shorted stock and replace it with a stock of the same style without violating the investment mandate.
The authors clarify the determinants of mutual fund performance and performance attribution, especially for family-fund strategists focused on asset allocation across different styles. They further refine not only the understanding of the impact of restrictions on fund performance but also models that assess the trade-off between maximizing family fund–level performance and maximizing fund-level performance.
How Did the Authors Conduct This Research?
The Investment Company Act of 1940 requires investment firms to file N-SAR reports covering 133 questions related to the investment practices of the firms’ funds. From the SEC’s EDGAR database, the authors gather the N-SAR reports filed between 1996 and 2008 for US open-end domestic equity mutual funds. They match each fund to the CRSP mutual fund database to collect data on performance and fund characteristics, excluding sector funds, enhanced index funds, and funds with no style category. The final sample contains 2,070 funds, of which 1,924 are active funds and 142 are index funds.
Dummy variables are used to indicate funds that lend securities and those that do not. A number of fund-level variables are used to calculate the value-weighted average short-interest and institutional ownership of the stocks in the fund’s portfolio. The authors construct a measure of fund manager investment restrictions for each year and a measure of whether the investment manager can invest outside the primary investment objective.
They then estimate panel regressions of monthly fund after-fee performance on the security-lending-used dummy variable and other fund characteristics. The dependent variable is the fund’s monthly four-factor alphas. The authors control for short-selling supply and demand and for such variables as expense ratio, total net assets, net flow, turnover, family total net assets, the portion of funds sold through family brokers, and past performance.
Finally, they estimate ordinary least-squares regressions of the Carhart four-factor alphas (in monthly percentages) on the security-lending-used dummy variable and lagged fund characteristics. The t-statistics indicate that the coefficient is significant at the 1%, 5%, and 10% levels. The results are consistent with the underperformance of lending funds being concentrated among active managers whose investment mandates restrict their investment choices.